Uncertainty Management Framework for Space-based Autonomy

Uncertainty management within autonomous systems is imperative to improving trust in multi-agent cyber-physical-human systems. These multi-agent systems require not only a robust uncertainty quantification strategy, but also an intelligible representation of uncertainty to human agents to elucidate confidence in the autonomous portions of the multi-agent systems. Here, we propose to develop a general framework for integrating uncertainty management within a variety of missions relevant to NASA. This framework will include key elements of deep learning for autonomous decision-making, quantification of data and model uncertainties, model explainability, and effective representation of system status to human agents. To achieve the goals of this project, CFD Research is partnering with the University of Michigan to transition recent advancements in the areas of deep learning, autonomy, uncertainty quantification, and mission management into an operational setting. In this program, space rendezvous will serve as the primary design reference mission for which to demonstrate the uncertainty management framework. In Phase I, we performed a proof-of-concept study where perception-based deep learning models were trained to perform pose estimation of a target spacecraft. These models were initially trained on publicly available spacecraft imagery datasets (e.g., SPEED+), then subsequently refined with renders of the Gateway outpost to align the work with NASAs current objectives. Uncertainty quantification and model explainability approaches were used to analyze the resulting model predictions. In Phase II, the technical approach will be further refined to include active learning and on-line model refinement, transfer learning, uncertainty-aware navigation and control, and a prototype user interface.   Uncertainty management within autonomous systems is imperative to improving trust in multi-agent cyber-physical-human systems. Here, we propose to develop a general framework for integrating uncertainty management within a variety of missions relevant to NASA. To achieve the goals of this project, CFD Research is partnering with the University of Michigan to transition recent advancements in the areas of deep learning, autonomy, uncertainty quantification, and mission management into an operational setting.   Innovations: 1) Uncertainty Quantification (UQ) framework for space missions; 2) Human representation of UQ to support multi-agent cyber-physical-human teams; 3) Demonstration and validation of the prototype software for mission planning at the Gateway outpost; 4) Modular and extensible software framework to support a variety of NASA-relevant missions beyond rendezvous with the Gateway outpost; 5) Technology insertion into existing autonomous, decision-making, and process control frameworks. The overall objective of the proposed work is to develop an extensible, modular framework for uncertainty management in NASA-relevant autonomous system applications. The specific objectives of this Phase II effort are focused on maturing the current capabilities of the uncertainty management framework whose feasibility was demonstrated during Phase I, while also demonstrating key new features, such as the use of the predicted uncertainty in downstream control and decision-making processes. Therefore, the specific objectives for Phase II are: Refinement of Gateway pose estimation model through additional scene generation using the DIRSIG framework; Develop a capability for active learning, where existing models may be refined through new data acquisition; Assess performance of transfer learning approaches to fine-tune existing spacecraft-specific models for estimating the pose of new spacecraft; Demonstrate the uncertainty management framework for more complex, real-world examples, including the presence of multiple imaging sensors on the chaser spacecraft; Demonstrate the potential for the quantified uncertainty in pose estimates to be used downstream within a GN&C algorithm; Document and transfer the resulting uncertainty management software to NASA for application.   Deliverables will include all developed software, and reports documenting methods and results.

This topic directly addresses NASA’s needs to address gaps in managing data and model uncertainties, especially for current and future activities by various mission directorates, including Space Technology Mission Directorate (STMD), Science Mission Directorate (SMD), and others. As a vital component of NASA’s Artemis program, the design reference mission of the proposed effort, rendezvous with the Gateway outpost, should allow for NASA to directly leverage this technology in future activities. An uncertainty management framework within the context of a multi-agent cyber-physical-human autonomous systems will be a valuable, cross-cutting capability in numerous domains. For example, terrestrial autonomous systems (e.g., autonomous vehicles) have similar challenges for uncertainty representation to humans and performing optimal decision-making which properly accounts for uncertain data.

This is early/mid-stage (TRL 5) — the most realistic path in is NASA SBIR/STTR, which funds small businesses and research institutions to develop technology aligned with NASA's needs (equity-free, phased funding). Check whether a current SBIR/STTR solicitation topic overlaps with this project's technology area, or contact the project directly (above) to ask.

None of these are guaranteed paths for this specific project — TechPort itself doesn't have an "apply" button. Reaching out to the contact(s) above with a specific question is usually the fastest way to find out what's actually open.